Report a specific diagnostic for walrus in a comprehension iterable expression

[Henry Su (hsusul)]

Summary

An assignment expression (:=) used anywhere within the iterable expression of a comprehension's for clause is a syntax error under PEP 572, and — unlike a walrus used bare as a comprehension if condition — it cannot be made legal by adding parentheses. Pyright already reports this as an error, but it reused the generic message:

Operator ":=" is not allowed in this context without surrounding parentheses

That message is misleading here: the offending walrus in the reproduction is already parenthesized, and no amount of parenthesization will fix it. This addresses #11514.

Minimal reproduction

result = [
    y
    for y in [
        z
        for k in range(10)
        if (z := k * 2) % 3 == 0   # inner comprehension is the outer iterable
    ]
]

CPython:

SyntaxError: assignment expression cannot be used in a comprehension iterable expression

Current behavior

Pyright reports Operator ":=" is not allowed in this context without surrounding parentheses, even though the walrus is parenthesized.

Corrected behavior

Pyright reports Operator ":=" is not allowed within a comprehension iterable expression.

Root cause

In parser.ts, _parseAssignmentExpression emitted walrusNotAllowed for two distinct conditions that were OR'd together:

if (!this._assignmentExpressionsAllowed || disallowAssignmentExpression) {
    this._addSyntaxError(LocMessage.walrusNotAllowed(), walrusToken);
}

These two flags actually represent different restrictions:

• !this._assignmentExpressionsAllowed is set only by _disallowAssignmentExpression, whose only caller wraps parsing of a comprehension for ... in <iterable> sequence expression. This is the "walrus in a comprehension iterable" restriction, which parentheses do not resolve.
• disallowAssignmentExpression corresponds to a bare walrus that requires surrounding parentheses (e.g. an unparenthesized walrus used as a comprehension if condition, or a bare walrus where a test expression is expected). Here parentheses do resolve the error, so the existing message is correct.

Implementation

• Split the condition so the comprehension-iterable case emits a new dedicated message walrusNotAllowedInComprehension, while the parentheses case keeps walrusNotAllowed. The iterable case is checked first so it wins when both apply (parentheses would not help there either).
• Added the walrusNotAllowedInComprehension key to localize.ts and package.nls.en-us.json.

Error detection (which cases are flagged, and the diagnostic range) is unchanged — only the message text differs for the comprehension-iterable case.

Tests

• New sample assignmentExprMessage1.py and test AssignmentExprMessage1 assert the exact message for both branches: the comprehension-iterable case gets the new message, and a bare walrus in a comprehension if keeps the generic message.
• The existing AssignmentExpr4 test (which already exercises iterable-walrus cases) continues to pass with its error count unchanged, confirming detection is unaffected.

Validation

Run in packages/pyright-internal:

• npx tsc --noEmit — passes.
• npx prettier@2.8.8 -c on all changed TS/JSON files — "All matched files use Prettier code style!".
• git diff --check — clean.
• npx jest parser tokenizer typeEvaluator1 — all pass (includes the new AssignmentExprMessage1 and the unchanged AssignmentExpr4).
• Full npx jest — all suites pass except LSP/type-server integration suites, which fail to run in this environment because they require the full monorepo bootstrap and the webpack test-server bundle (Cannot find module 'fs-extra' / Server bundle does not exist); these are unrelated to this change.

Compatibility

No API changes. The only behavioral difference is a more accurate diagnostic message for walrus operators inside a comprehension iterable expression; the set of reported errors and their source ranges are unchanged.

[Stella Huang (StellaHuang95)]

🔒 Automated review in progress — Stella Huang (@StellaHuang95) is auto-reviewing this PR.

[Stella Huang (StellaHuang95)]

Approved via Review Center.

[Rich Chiodo (rchiodo)]

Approved via Review Center.

[Rich Chiodo (rchiodo)]

Approved via Review Center.

[Stella Huang (StellaHuang95)]

Approved via Review Center.

[Rich Chiodo (rchiodo)]

🔒 Automated review in progress — Rich Chiodo (@rchiodo) is auto-reviewing this PR.

[Rich Chiodo (rchiodo)]

Approved via Review Center.

[Heejae Chang (heejaechang)]

Approved via Review Center.

[Rich Chiodo (rchiodo)]

Info · Optional note

Consider adding the issue's nested-comprehension reproduction to this exact-message test. Existing tests cover nested detection, but not the new message for that precise shape.

[verified]

[Rich Chiodo (rchiodo)]

Verification: The relevant tests could not be fully run in the isolated environment; this review is not fully verified.